Open Collections

Description

This is the Mandarin–English Language Interview (MELI) corpus. MELI is an open-access audio corpus of Mandarin–English bilingual speech recorded in Vancouver, Canada in 2024. The corpus contains approximately 30 hours of interview speech from 51 Mandarin–English bilingual speakers, who completed sociolinguistic interviews and read sentence tasks in both languages, with a draw-a-map activity in their Mandarin interviews. For each speaker and language, high-quality audio recordings are accompanied by Praat TextGrid files that include orthographic transcriptions at utterance-, word-, and segment-level alignments. The corpus also includes digitized maps from the draw-a-map task, a tabular data file of participants' language background, two text files of out-of-vocabulary (OOV) items manually added in forced alignment, and a detailed README file describing corpus structure, annotation procedures, and metadata. As an open-access resource, MELI supports research in sociophonetics and bilingualism, and enables the study of how social factors interact with acoustic detail in bilingual speech. Corpus documentation is available at: https://meli-corpus.readthedocs.io/en/latest/.